Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
WestminsterResearchhttp://www.westminster.ac.uk/westminsterresearch
Post-Brexit trade survival: Looking beyond the European Union
Jackson, K. and Shepotylo, O.
NOTICE: this is the authors’ version of a work that was accepted for publication in
Economic Modelling. Changes resulting from the publishing process, such as peer
review, editing, corrections, structural formatting, and other quality control mechanisms
may not be reflected in this document. Changes may have been made to this work since
it was submitted for publication. A definitive version was subsequently published in
Economic Modelling.
The final definitive version in Economic Modelling is available online at:
https://doi.org/10.1016/j.econmod.2018.04.010
© 2018. This manuscript version is made available under the CC-BY-NC-ND 4.0 license
http://creativecommons.org/licenses/by-nc-nd/4.0/
The WestminsterResearch online digital archive at the University of Westminster aims to make the
research output of the University available to a wider audience. Copyright and Moral Rights remain
with the authors and/or copyright owners.
Whilst further distribution of specific materials from within this archive is forbidden, you may freely
distribute the URL of WestminsterResearch: ((http://westminsterresearch.wmin.ac.uk/).
In case of abuse or copyright appearing without permission e-mail [email protected]
1
Post-Brexit trade survival: Looking beyond the European Union
Abstract
As the EU and UK negotiate a new relationship, this paper explores the welfare implications of this policy change and its interaction with major trade policy initiatives. We evaluate five Brexit scenarios, based on different assumptions regarding Brexit, TTIP and various free trade deals the UK may attempt to broker with the US or Commonwealth countries. We also consider the dynamics of welfare changes over a period of two decades. Our estimates suggest that the impact of Brexit is negative in all policy scenarios, with lower welfare losses under a soft Brexit scenario. The losses are exacerbated if TTIP comes into force, demonstrating the benefits of being a member of a large trade bloc. However, they occur gradually and can be partially compensated by signing new free trade agreements. To further minimise losses, the UK should avoid a hard Brexit.
JEL codes: F13, F15, F17Keywords: Brexit, Transatlantic Trade and Investment Partnership, Gravity, UK, EU
1. Introduction
During the last century, the UK has transitioned from global trade power to
‘awkward partner’ in the EU; then most recently an outright rejection of EU
membership. Following these events, a new wave of academic literature
exploring the potential impact of Brexit has started to emerge (Ottaviano et
2
al., 2014; Dhingra et al., 2016; 2017, Ebell et al., 2016).1 These nationalist
tendencies are in contrast to the moves towards forming mega-regionals,
where the EU negotiating team have been provided with a mandate to
negotiate the Transatlantic Trade and Investment Partnership (TTIP) with
the US.2 Therefore, research exploring the effects of TTIP has also started
to emerge (Francois et al., 2013; Egger et al., 2015; Felbermayr et al., 2015;
Disdier et al. , 2016).3 For the UK, its withdrawal from the EU is expected to
exclude it from any TTIP deal. In addition, there have been mixed messages
about the likelihood of restarting the TTIP negotiations and if the EU would
be in front of the UK (US-UK FTA) in the queue.
Our contribution is to cut across these separate strands of literature and
in doing so we assess the UK position in terms of TTIP and Brexit scenarios.
We also go further than the existing Brexit literature by modelling future
Commonwealth trade relationships as well as a possible US-UK FTA after
Brexit. The main channel of welfare effects that we consider comes from
the general equilibrium effect of changes in trade costs. Our analysis
adopts an emerging research framework, which allows us to predict the
impact of policy changes at the extensive and intensive margins, where this
two-step methodology has not been applied within the Brexit literature.
Given the uncertain outcomes and lack of detail regarding potential policy
1 Dhingra et al. (2017) estimate welfare losses of 1.3 percent in a static model with ‘soft Brexit’ compared to 9.4 percent in a dynamic model with ‘hard’ Brexit. On the other hand, Ebel et al. estimate that real income will fall by 2.7 percent in the long run. For broad and comprehensive summary of Brexit literature, including future UK trade policy options and Brexit consequences for the UK and EU, please see Sampson (2017).2 The term mega-regionals refers to the TTIP, TPP and Regional Comprehensive Economic Partnership (RCEP).3 Felbermayr et al. predict gains of 3.9% for the EU, where Francoise et al. estimate a more moderate 0.48%. Furthermore, Felbermayr et al. expect third countries, outside the TTIP, to lose 0.9% due to trade diversion. Egger et al. suggest that the negative impact on third countries is likely to be exacerbated by a more discriminatory TTIP agreement. Disdier et al. investigate the interplay of the TTIP and Trans-Pacific Partnership (TPP) in terms of agriculture.
3
changes concerning TTIP and Brexit, it is problematic to pin down the
changes in tariffs and non-tariff measures. We deal with this issue by
considering ‘soft’ and ‘hard’ Brexit scenarios, which are evaluated based on
actual data for trade effects of regional trade agreements (RTA) in general,
and customs unions (CU) and free trade agreements (FTA) in particular, in
162 countries over the period 1960-2014. We also look at the dynamics of
RTA effects on trade, conditional on its duration and decade, then
incorporate it in our welfare analysis. Our underlying assumption is that
effects of future trade agreements can be inferred from the effects of
existing trade agreements. If, as some would argue, the TTIP goes further
than the previous RTAs in terms of lowering trade costs and reducing waste
of resources due to harmonization of actionable NTMs (see Berden et al.,
2009), our baseline estimates would indicate a conservative lower bound.
This paper will highlight the losses stemming from the lower bargaining
power of the UK as a stand alone country, rather than as a member of a
powerful trading bloc.4 We will also highlight that signing FTAs with more
remote countries, such as the US or Commonwealth countries (Australia,
Canada, New Zealand, India, Pakistan, South Africa) cannot compensate for
these losses, as some commentators suggest. In case of a ‘hard’ Brexit, the
TTIP would generate additional welfare losses for the UK equal to 0.5-2
percent of GDP. On the other hand, our results for the soft Brexit scenario
indicate that the UK would experience welfare losses of approximately 3.1
percent if the TTIP comes into force alongside the UK withdrawal from the
EU. The negative effect of Brexit alone is 2.7 percent. Signing an FTA with
4 “The US government has proposed slapping punitive tariffs on Bombardier, a Canadian aircraft maker with operations in Northern Ireland. … The Bombardier case gives the lie to the notion that, outside the EU, the UK will find the behaviour of its leading non-European trade partners more benign.” Financial Times, “Bombardier exposes post-Brexit realities”, September 28, 2017 by Tony Barber.
4
the US or large Commonwealth countries would lower welfare losses by
approximately 0.65 to 1.1 percent, depending on the scenario. A ‘hard’
Brexit would generate losses from 4.1 to 5.3 percent. These results are
robust to various levels of TTIP impact on trade (as long as they are
positive), external effects of higher economic growth caused by TTIP, and
alternative definitions of Brexit.
We also run a number of robustness checks. First, we present evidence
of heterogeneous effects of different types of RTAs on trade, which is
related to extent that NTMs go beyond tariffs. Second, we consider how
additional waste of resources due to NTMs and border controls influences
welfare (see Baldwin et al., 1997).5 Finally, we perform a sensitivity analysis
by changing our assumptions about the trade creating effect of TTIP and its
impact on economic growth.
There are two counteracting tendencies that should be taken into
account when interpreting welfare gains. First, even the ‘hard’ Brexit
scenario would not cause abrupt welfare losses and the catastrophic decline
of trade flows between the EU and UK. These changes would occur
gradually. However, RTAs (especially CUs and the EU in particular) had the
strongest trade creating effect over the last 50 years, which would indicate
that the welfare losses from Brexit are likely to be higher and the welfare
gains of TTIP would increase over time.
The rest of the paper is organized as follows. Section 2 presents an
overview of the evolution of the multilateral framework and the mega-
regionals in the context of the UK. Section 3 discusses potential trade policy
5 “Whitehall is planning to hire another 2,000 staff to deal specifically with Brexit in a sign of how its resources are being diverted towards the challenges of leaving the EU.” Financial Times, “Ministers haggle over 2,000 new staff as Brexit tests civil service”, October 12, 2017, by Jim Pickard and George Parker.
5
scenarios. Section 4 outlines the methodology and data, where Section 5
discusses the results of the analysis. Section 6 performs robustness checks.
Finally, Section 7 concludes.
2. Multilateral negotiations and mega-regionalsHistory plays an important role in determining trade (Eichengreen and Irwin,
1998). Krasner (1976) emphasised the role of past decisions and policies,
with particular reference to the UK and US as architects of the international
trade structure. In the aftermath of World War II (WWII) the trade
environment featured entrenched protectionist positions, following the
market failure of the inter-war years (Irwin, 2011). The best case of global
free trade, as proposed by neoclassical trade economists, was infeasible.
Hence, national objectives including economic growth and political power
were key issues. By the end of WWII, the UK’s position as the dominant
global trade power was overtaken by the US. During this period, there was
a broad acceptance of the ‘trade promotes peace’ hypothesis (Anderton and
Carter, 2001; Hull, 1948). Therefore, these two powerful states attempted
to balance their interests in terms of political strength and a desire for global
stability. Hence, reeling from conflict, the states looked to reduce trade
barriers. However, perhaps inevitably, they remained unable to ignore
national interests (Mavroidis et al., 2009).
Multilateralism held considerable appeal for policy makers looking to
reverse the economic failures of the inter-war period. This was both in terms
of promoting peace as well as swifter liberalisation than what was likely
from negotiating separate bilateral deals. The US and UK policy makers
dominated discussions that led to the signing of the GATT, the precursor to
6
the WTO (Toye, 2003). A range of concerns, including imperial preferences
and promotion of the capitalist model, promoted legitimising FTAs and CUs
within the GATT (Article XXIV, GATT). Despite the concessions regarding
imperial preferences, from the 1950s, UK trade reoriented away from their
(former) colonies towards European trade partners (Anderson and Norheim,
1993). More generally, Article XXIV permitted the segmentation of the world
into overlapping trading blocks, where almost every country is a member of
at least one RTA. Multilateral negotiations initially focussed on tariffs then
as the attention moved towards NTBs, and regulation in particular,
agreements proved more difficult. Low (2015) argued that the Article XXIV
concession was pivotal. Therefore, this original compromise is key to
understanding the current issues facing the WTO.
Recently, the WTO has struggled to provide an effective multilateral
negotiating framework (Azevêdo, 2016; Froman, 2015; Wilkinson, 2002).
This vacuum and the emergence of the BRICS (Brazil, Russia, India, China
and South Africa) countries as key players in the global economy provided
fertile ground for the development of the mega-regionals (Figure 1). These
new groupings provide a much greater challenge to the WTO than their
predecessors, which were geographically localised and considerably
smaller. Furthermore, the mega-regionals are looking to address regulatory
barriers, which have proved difficult to deal with on a multilateral basis. The
WTO’s inability to meaningfully regulate RTAs, where the seed was sown
during the inception of the GATT, suggests that the WTO will not block the
formation of the mega-regionals. Moreover, there are suggestions that the
WTO may view mega-regions as a stepping-stone towards multilateral
liberalisation (Schott, 2014).
7
Negotiating these blocks requires large-scale trade diplomacy and
prolonged discussions. This makes the US’s recent withdrawal from the TPP
and suspension of the TTIP talks particularly unwelcome. Since WWII, the
key players have shifted from the US and UK to the US and China. The US
is witnessing a decline in its position, not unlike the UK in the early 20th
century. Where the UK attempted to retain its strong position by arguing for
imperial preferences, similarly, the US is looking to its historical
partnerships (and potentially, mega-deals) to tackle the growing strength
of China. However, we should not underestimate the importance of the EU
trading bloc, where Fan et al. (2014) use network analysis to find that
disconnecting EU exports from Japan or the US would lead to the collapse
of the world trade system. Therefore, the mega-regionals/blocs have
formed an important part of the geopolitical ‘game’. These mega
arrangements are likely to impose rules on those countries outside the
blocks, suggesting that been outside one of the three clubs could prove
costly (Li et al., 2016).
The EU also shares the US concerns regarding the growing
importance of China. They have been engaged in extensive negotiations
towards a new China-EU Investment Agreement, which may turn into a
China-EU FTA. Furthermore, EU members fear a US-Asia powerhouse from
which they would be excluded. The TPP was signed 4 February 2016,
providing an incentive for the TTIP negotiations to forge ahead (until
recently). The main points of contention include agricultural subsidies,
services (i.e. audio and visual, financial), food safety and environment,
labour standards, investor-to-state dispute settlement system and public
procurement. EU and US levels of GDP and trade make this a formidable
arrangement (Akhtar and Jones, 2014). However, the process has been
8
plagued with controversy, leaving the Commission to insist that they are
engaging with the public as openly as possible (Malmström, 2015).
Supporters of the TTIP claim that it may raise welfare in the EU and US
through increases in productivity and higher consumer surplus. Opponents
argue that the welfare gains would be small, while the effects on standards
in the labour market, health and safety in product markets, and
environmental standards may outweigh the benefits (Pitlik, 2016).
Figure 1: Mega-regionals
Source: Authors work
Notes: 1/ TPP: countries only included in this mega-regional arrangement 2/ TTIP: countries only included in this mega-regional arrangement 3/ TPP and TTIP: countries included in both arrangements 4/ RCEP: countries only included in this mega-regional arrangement 5/ TPP and RCEP: countries included in both arrangements
Meanwhile, UK trade negotiations have been integrated and subsumed
within the EU. EU member states, including the UK, monitor and assist the
EU Commission but are not permitted to conduct independent trade
negotiations. Whereas, trade missions to extra-EU countries involving
business-to-business (B2B) dialogues are targeted at internationalising
firms. Therefore, Brexit presents the UK government with significant
9
challenges including increasing capacity within the Department for
International Trade (previously UK Trade and Investment, until July 2016).
This newly formed department will need to address rising costs through
trade preference erosion as well as the UK being a less attractive
destination for investment. This situation has been further exacerbated by
the recent UK election results, which provide an even more questionable
basis for exit negotiations. In the face of such difficulties it is somewhat
predictable that the Commonwealth is mentioned. The Commonwealth
remains but re-establishing it as a strong preferential trading club headed
by the UK, or the UK signing of a collection of bilateral FTAs, may prove
difficult (Figure 2). After WWII, the UK defence of imperial preferences was
seen by some as a desperate attempt to maintain a powerful position in the
world-trading environment. In fact, it was increasingly the case that the UK
relied on the colonies, particularly during WWII (Lee, 1977). More recently,
the situation has worsened; with African countries becoming reticent to sign
up to EU trade deals as well as accusations of such deals being a type of
colonialism6. At the same time, India is backing their ‘Buy in India’
programme rather than looking to a new deal with the UK7. In summary, the
relevance of Commonwealth nostalgia for post-Brexit UK trade deals is
questionable. Therefore, the UK is struggling to retain a powerful position in
the world-trading environment. A key feature of this power-play is the TTIP
and the possibility of a US-UK FTA. At the same time there is the complex
and emotive issue of the UK relationship with the EU, whether this is a ‘soft’
or ‘hard’ Brexit.
6 On 26 Feb 2017, Tanzanian President John Magufuli held a joint press conference with visiting Ugandan President Yoweri Museveni where he described the Economic Partnership Agreement with the EU as a form of colonialism.7 Public Procurement (Preference to Make in India), Order 2017 was issued on 15 June 2017
10
Figure 2: Members of the commonwealth
Source: Authors work
3. Trade policy scenarios: interaction of mega-regionals with other agreements and impact on third countries
By lowering tariffs and eliminating non-tariff measures, bilateral and
plulateral trade agreements increase national welfare due to allocation and
accumulation effects (Baldwin & Venables, 1995). We mainly focus on the
effect of redistribution of trade caused by the interplay of TTIP and Brexit,
but briefly consider the effect of changes in resource costs due to
elimination/introduction of NTMs (Baldwin et al., 1997).
Once countries join the TTIP by virtue of EU membership, third countries
are expected to experience from trade diversion and trade preference
erosion. Countries outside the EU will find it difficult to negotiate similar or
better terms. Furthermore, if mega-regional blocs expand their membership
to new countries this could generate increased costs of being outside the
EU. We examine a number of scenarios in terms of the UK relationship with
the EU and the outcome of the TTIP negotiations:
11
1. EU&TTIP: UK remains a member of the EU and the TTIP is signed
(Scenario 1)
2. Brexit&TTIP: UK exits the EU and the TTIP is signed (Scenario 2)
3. Brexit: UK exits the EU and the TTIP is not signed (Scenario 3)
In addition, we consider two post-Brexit scenarios, which are discussed as
alternative UK policies in order to mitigate consequences of Brexit:
4. UK&US: UK exits the EU and successfully negotiates an FTA with the
US (Scenario 4)
5. Commonwealth FTA: UK exits the EU and signs FTAs with the largest
Commonwealth countries: Australia, Canada, India, Nigeria, New
Zealand, Pakistan, and South Africa (Scenario 5)
The last two scenarios are chosen to investigate the validity of a popular
claim that the UK has an opportunity to compensate the losses in the
European markets by freely negotiating FTAs with non-European countries.
We model the UK exiting the EU under soft Brexit and ‘hard’ Brexit
scenarios. Under the ‘hard’ Brexit, the UK would not be able to maintain
preferential tariffs with the EU, and will instead face the Most-Favoured
Nation (MFN) rates. We also assume that the NTBs applied to UK exporters
would be similar to those applied to exporters from outside the EU free trade
zone. Furthermore, in terms of the TTIP, we assume that the effect of the
arrangement would be typical of the FTAs signed over the period 1960-2014.
Therefore, to model the effect of the TTIP we rely on our estimates of the
average effect of RTAs during 1960-2014.
4. Methodology and DataThe underlying model for our analysis is a structural ‘new trade theory’
model (Helpman et al., 2008), which captures selection into positive
12
bilateral trading partners and the effect of trade policy on extensive and
intensive margins of trade. As pointed out by Head and Mayer (2014), the
structural gravity model is consistent with a wide class of models, including
Armington (1969), Krugman (1980), and Melitz (2003). This model has been
used to estimate the effect of preferential trade agreements (Egger et al.,
2011).
Trade flows are determined by
(1)Xij =YiEj
ΩiPjφij
where is exports from country i to country j, is total production Xij Yi = ∑jXij
in country i and is total expenditure in country j. is Ej = ∑iXij Ωi = ∑jφijEj
Pj
outward multilateral resistance and is the inward multilateral Pj = ∑iφijYi
Ωi
resistance term.8 Furthermore, represents bilateral trade costs, which are φij
influenced by the choice of trade policy. Our bilateral trade costs are given
by
(2)ln φij = ln distij + EUij + RTAij + Zijγ + uij
where is the logarithm of bilateral distance between trading partners lndistij
i and j, is a dummy variable representing EU membership status, EUij RTAij
is a dummy variable representing the presence (or otherwise) of an RTA and
denotes other variables influencing trade costs (e.g. common language, Zij
common legal system, colonial relationship).
8 When a large economy or trade block changes its policy towards another country, it influences bilateral trade directly and the trade of other countries indirectly. However, the indirect effect can be large. For instance, Russia introduced an embargo on imports of food from the EU in 2014. This action was in response to economic sanctions and it lead to the growth of food imports from countries that did not support the sanctions. Anecdotal evidence suggest that this policy allowed countries such as Belarus, which is landlocked, to export shrimps and salmon to Russia. At the same time, sales of Norwegian salmon to Belarus increased by 50 percent, where a large proportion was re-exported to Russia under new label of “Belorussian” salmon (Newsweek, “Busting a few lox”, 19, 09/04/2015).
13
Our empirical strategy proceeds in two stages. The first stage involves
modelling the selection into trade partners as well as estimating the trade
elasticities with respect to trade policy variables. To estimate extensive and
intensive trade elasticities with respect to trade policies, we follow Egger et
al. (2011). This procedure allows us to estimate a structural model with
responses to trade policy at extensive and intensive margins. It also
addresses the issue of zero trade flows and provides unbiased estimates in
the presence of heteroskedasticity. The second step uses simulations by
means of a structural gravity approach to model the alternative
scenarios/counterfactuals outlined above. We compute welfare gains/losses
of each of the scenarios compared to the status quo. Therefore, we examine
changes in real income using results derived by Arkolakis et al. (2012) and
the structural gravity equation (1).
4.1. Modelling the selection into trading partners and estimating trade policy effects
We first deal with the sample selection correction. Some countries are more
likely to trade than others and the probability of positive trade flows
depends on fixed and variable trade costs. Therefore, we estimate a probit
model to explore , the probability of positive trade flows between i and Tij,t
j:
(3)Tij,t = Pr (Tradeij,t = 1│.) = Φ(Hij,tΓTrade)
where is a binary variable that takes value of 1 if we observe Tradeij,t
positive trade flows and zero otherwise, denotes policy variables (RTA Hij,t
and EU) as well as the determinants of fixed and variable trade costs
including distance, common border, common spoken language, common
legal system and year fixed effects. The exclusion restriction that we exploit
is common religion, which is customary in the literature. We further form
14
variables that control for the selection into positive trade flows and
heterogeneity of exporting firms (Helpman et al., 2008):
(4)ρij,t = Φ(Hij,tΓTrade) = Φ(zij,t)
(5)ηij,t =ϕ(zij,t)Φ(zij,t)
(6)zij,t = zij,t + ρij,t
Next, we augment the gravity equation (1), by using , a third degree η
polynomial of , as well as introducing a full set of exporter-time, importer-z
time and bilateral fixed effects to control for multilateral resistance terms:
lnXij,t = γ0 + γ1EUij,t + γ2RTAij,t + γ3ηij,t + ∑3m = 1γm + 3(zij,t)
m + Dit + Djt + Dij (7)+ ϵij,t
We estimate (7) using a panel of bilateral export and import data from 1960-
2014 for 162 countries. The length of the sample and larger cross-section
of countries distinguishes our estimation from both Felbermayr et al. (2015)
and Dhingra et al. (2017). It gives us more time and cross-section variation
in order to pin down the effects of RTAs in general and EU in particular on
trade. It also gives us an opportunity to look at dynamics and trends in the
effect of RTAs on trade. We overcome the computational issue of dealing
with 36,000 fixed effects by applying an algorithm developed by Guimaraes
and Portugal (2010).
4.2. Accounting for endogenous RTA formation
The outlined procedure captures heterogeneity of country-pairs, controls for
country-time specific effects and accounts for the selection into trading
partners. However, it does not deal with endogeneity of trade policy. A
decision to sign an RTA is driven by bilateral relationships between
countries that may evolve over time. Moreover, it is influenced by trade
costs. Ignoring the selection into RTA partners is likely to bias downwards
15
the estimation of the effect of an RTA on trade (Baier and Bergstrand,
2007).9 We deal with this issue by modelling the selection into RTA and EU
partners as well as further instrumenting our policy variables with obtained
selection probabilities. We estimate the probit models of RTA and EU
formation as follows:
(8)δRTAij,t = Pr (RTAij,t = 1│.) = Φ(Gij,tΓRTA)
(9)δEUij,t = Pr (EUij,t = 1│.) = Φ(Gij,tΓEU)
where captures fixed and variable trade costs including distance, Gij,t
common border, common spoken language and common legal system. In
addition, for (9) we include an indicator variable, whether both countries are
located in Europe since joining EU is only allowed for European countries.
We then re-estimate model (7) using instrumental variables method, where
RTA and EU variables are instrumented by predicted values of and δRTAij,t δEU
ij,t
and the inverse Mill’s ratios.
4.3. Simulations
Once we have estimated the RTA and EU impact on trade, we start the
simulation part of our analysis. Following Anderson et al. (2015), we
estimate a structural gravity model using 2012 data and applying the
Poisson Pseudo Maximum Likelihood estimator (PPML estimator, see Silva
and Tenreyro, 2006). We constrain the coefficients of the policy and
selection variables to be equal to our estimated coefficients from the
previous stage. Our estimated model is given by
Xij = exp (γ1EUij,t + γ2RTAij,t + γ3η + ∑3m = 1γm + 3(zij,t)
m + Zijπ + χi + ξj) + (10)υij
9 The downward bias occurs if high, unobserved bilateral trade costs makes an RTA more likely.
16
where are bilateral trade costs variables including distance, common Zij
border, colonial links, common legal origin, common spoken language and
common religion. Using result by Fally (2015), we compute the inward and
outward multilateral resistance terms according to the following
expressions:
(11)P1 - σj = Ejexp ( - ξj)/E0
(12)Ω1 - σi = E0Yiexp( - χi)
where is the level of expenditure in the country for which the inward E0
multilateral resistance is normalized to .10P0 = 1
The model is further re-estimated with modified policy variables EU *ij,t
and , which switch on and off according to various scenarios, as well RTA *ij,t
as the inward and outward multilateral resistance terms re-calculated with
new fixed effects, and , according to the expressions (11) and (12).χ *i ξ *
j
4.3.1. Conditional Effects of TTIP and Brexit
Conditional effects of a change in policy, (i.e. the UK exiting the EU or the
EU signing the TTIP) are modelled by modifying policy variables to reflect
the expected policy changes, taking the income and expenditure of
countries, and , as given.11 This approach highlights the changes in Yi Ej
trade patterns due to the changes in the corresponding trade costs caused
by policy changes. Therefore, it does not rely on any other channels of
economic changes required for gains/losses to materialize. The limitations
of this approach are as follows. First, it assumes that the effect of the TTIP
is equal to the average effect of the RTAs in the past. Second, it shuts down
10 In our estimation, Afghanistan is the reference country. Welfare changes are estimated relative to Afghanistan.11 Head and Mayer (2014) label this approach as Modular Trade Impact (MTI).
17
the effects that the TTIP and UK exit from the EU may have on production,
technology, employment and relative wages. Third, it does not account for
welfare gains caused by saving resources due to NTM removal. Fourth, it
does not account for the effect on non-tradables and services. We further
discuss and address these issues in the robustness checks section.
4.3.2. Evaluating welfare
We compute changes in direction and value of trade and evaluate welfare
changes according to the following:
(13)WCEi = 100% × (
Y *i
P *i
YiPi
- 1) = 100% × (Pi
P *i
- 1)
where are policy scenarios described in the previous section. i = 1,2,3,4,5
The last equality is due to the fact that in the conditional scenario we keep
outputs and expenditures constant.12
4.4. Data
Our dataset covers 162 countries for 1960-2014. Aggregate bilateral
exports and imports measured in thousands of current US dollars are taken
from the Direction of Trade Statistics (DOTS) provided by the International
Monetary Fund (IMF). To compute welfare gains, data on internal trade is
required. We approximate internal trade as given by
Xii = (1 - sservices) × GDPi - ∑j ≠ i
Xij + ∑j ≠ i
Xji
12 It is worth mentioning that our welfare results does not depend on the choice of the reference country. Since we keep expenditure levels of all countries constant, welfare changes are given by the ratio of exponentiated fixed effects in the status quo and
counterfactual scenarios: . Also, this formula is consistent with a wide class Pi
P *i
=exp ( - ξi)
exp ( - ξ *i )
of models, including a Melitz model with heterogeneous firms, which was the basis for our first-stage estimation.
18
where is the share of services in GDP, is total export of sservices ∑j ≠ iXij
country i, and is total import of country i.13∑j ≠ iXji
RTA data are taken from the Centre D'Etudes Prospectives et
D'Informations Internationales (CEPII) database and updated to 2014 using
the WTO database of regional trade agreements. RTA is a binary variable
that takes value of 1 if a country-pair has an active RTA agreement in place
(including the EU agreement) and 0 otherwise. In addition, we introduce a
binary variable ‘EU’ that takes value of 1 if both trading countries are EU
members and 0 otherwise. A positive and significant coefficient of the EU
variable would indicate that the EU generates more trade relative to the
other trade agreements. We complement the RTA data with Mario Larch’s
Regional Trade Agreements Database from Egger and Larch (2008). This
data allows us to distinguish between different types of RTAs, such as FTAs
and CUs. Finally, data on applied tariff rates is collected from the TRAINS
database for the period 2003-2013.
The data on Gross Domestic Product (GDP) in current US dollars and
total population are from the World Development Indicators (WDI) 2014
published by the World Bank. Geographical characteristics and distance
between countries are taken from CEPII, (see Head et al. (2010) for a
detailed description of the data). Colony and contiguity dummy variables
are used to control for pair-specific trade costs that are not directly related
to distance. Furthermore, the dummy representing common legal origin
captures the compatibility of the legal systems of trading partners and
captures trade costs related to the signing of contracts. The common
13 Data on the share of services in GDP in 2012 is taken from the World Bank. In several cases, when the data for 2012 is not available, the latest available data is taken.
19
spoken language and common religion dummy variables capture the effect
of cultural similarities on trade (Melitz and Toubal, 2014).
We carefully record both positive and zero trade flows for both exports
and imports, while distinguishing zero trade flows from missing data. Some
countries, such as the USSR, ceased to exist within this period. Our dataset
does not contain the bilateral trade of these countries. However, trade of
newly created states, such as Russia and Ukraine, is recorded since the date
of their creation.
5. Empirical ResultsWe present our empirical results according to the sequence of the
discussion outlined in the previous section.
5.1. Selection into positive trade and RTA partners
We begin with the selection into trade and preferential trading partners.
Table 1 reports estimates of these selection processes, where columns (1)
and (2) refer to the selection into exporting while controlling for RTA/EU and
FTA/CU, respectively. Columns (3) and (4) explore the determinants of the
selection into regional trading/economic blocks, while columns (5) and (6)
look at the differences in selection between FTA and CU.14 For each model
we report marginal effects estimated using the average values of the
explanatory variables. We find that an RTA is positively associated with
probability of exporting and importing. An RTA increases the probability of
positive exports by 17 percent. This aligns with our expectations given that
one of the main goals of trade agreements is to ease market access.
14 For instance, Lake & Yildiz (2016) argue that over longer distances FTA is preferable to CU.
20
Moreover, EU membership is positively associated with the probability of
trade, but the effect is not significant. The effect is more pronounced for a
CU, while an FTA increases probability of trade by only 9.8 percent. This
may reflect the fact that a CU goes beyond reductions in applied tariffs and
eliminates NTMs as well, which in the era of low tariffs play a more
prominent role in shaping trade flows.
Positive trade is more likely if countries share a common spoken
language and a common colonial past. Higher levels of production in
exporting countries and a higher level of expenditure in importing countries
are also positively linked to the probability of trade. On the other hand,
higher levels of population are negatively linked to probability of trade,
which among other things reflects a home market effect. We also find a
negative effect of common religion on trade (conditional on controlling for
common language, colonial past, etc.), which may be capturing the fact that
more similar countries have less incentive to trade.
According to the model in column (3), the probability of creating an RTA
is positively linked to a common spoken language and the economic size of
exporting and importing countries. Yet, it is negatively linked to a common
border, sharing the same religion and a larger population. The probability
of EU membership in column (4) is positively related to a common legal
system and larger economic size. The effect of economic size on being a
member of the EU for both exporting and importing countries is about four
times larger than the effect of forming a common RTA.
21
Table 1: Selection models
Selection into bilateral trading
partners
RTA and EU selection
FTA and CU selection
RTA/EU FTA/CU RTA EU FTA CU(1) (2) (3) (4) (5) (6)
RTA .170**
(.011)EU .073
(.042)FTA .098**
(.009)CU .184**
(.010)Both in Europe .013**
(.001)Com. border .024 .020 -.003* -.007** -.007** -.004**
(.022) (.021) (.002) (.002) (.002) (.001)Colonial past .165** -.002 -.004* .006* -.002
(.022) (.002) (.001) (.002) (.001)Com. legal .006 .007* .001 .002* .000 .000
(.003) (.003) (.001) (.001) (.001) (.000)Com. religion -.049** -.050** -.009** .001 -.000 -.003**
(.006) (.006) (.001) (.001) (.001) (.001)Com. language .269** .268** .016** .003 .024** .010**
(.007) (.007) (.001) (.001) (.001) (.001)ln Distij -.116** -.111** -.021** -.008** -.020** -.012**
(.002) (.003) (.001) (.001) (.001) (.001)ln GDPit .105** .106** .003** .006** .006** -.001**
(.001) (.001) (.000) (.000) (.000) (.000)ln Popit -.017** -.018** -.002** -.004** -.004** .001**
(.001) (.001) (.000) (.000) (.000) (.000)ln GDPjt .081** .082** .003** .006** .006** -.001**
(.001) (.001) (.000) (.000) (.000) (.000)ln Popjt -.009** -.010** -.002** -.004** -.004** .001**
(.001) (.001) (.000) (.000) (.000) (.000)Observations 965973 965973 965973 965973 965973 965973
Notes: ** Significant at the 1% level. * Significant at 5% level. Marginal effects estimated at average values are reported. Standard errors clustered at country-pair in brackets. All models are estimated by probit. Columns (1) and (2) report selection into exporting country-pairs. Columns (3) and (4) report selection into RTA/EU country-pairs. Columns (5) and (6) report selection into FTA/CU country-pairs. All models are estimated for 162 countries in 1960-2014.
22
FTA is more likely to be created between countries sharing colonial past,
while it is not a significant determinant of CU formation (columns (5) and
(6)). Common language is important for all types of RTAs, but is relatively
more important for FTA formation, while common religion has a negative
and significant effect on CU formation.
Table 2: Long-term gravity estimates
RTA and EU FTA and CU
Applied tariff
FE Selection
IV PPML
(1) (2) (3) (4) (5) (6)RTA .353** .562** 1.242** .100** -.034
(.026) (.053) (.103) (.028) (.059)EU .656** .630** .582** .382** .437**
(.040) (.043) (.110) (.035) (.053)FTA .296**
(.037)CU .928**
(.065)Applied tariff
-1.025**
ln(1 + tijt) (.136)ηijt 2.497** 2.625** 1.357**
(.120) (.121) (.264)zijt 3.455** 3.384** 3.171**
(.228) (.232) (.465)z2
ijt -1.242** -1.253** -.856**
(.074) (.074) (.145)z3
ijt .130** .135** .078**
(.008) (.008) (.015)Observations
579989 537390 537390 971189 537390 266554
R2 .864 .870 .868 .992 .871 .906Notes: ** Significant at the 1% level. * Significant at 5% level. Standard errors clustered at country-pair in brackets. All models are estimated with a full set of country-pair, exporter-year, and importer-year fixed effects. To deal with high dimensionality problem of adding more than 36,000 fixed effects, we use the algorithm developed by Guimaraes et al. (2010). Dependent variable in all models except (4) is ln(exp). Dependent variable in (4) is exp. Models (1)-(5) are estimated for 162 countries in 1960-2014. Model (6) is estimated for 162 countries in 1993-2014.
5.2. Long-run RTA elasticity of trade
Table 2 presents estimates of the effects of trade policy on exports at
intensive margins. We control for exporter-year, importer-year, and
country-pair time invariant effects. Standard errors are clustered by
23
country-pair. Columns (1)-(4) present results for all types of RTAs. In column
(1) we do not control for selection and heteroskedasticity driven by firm-
level heterogeneity. In this case, on average, signing an RTA increases trade
by 35.3 percent. Joining the EU will have even stronger trade-creating
effect, additionally boosting exports by 65.6 percent. These numbers are
slightly higher than are typically found in other studies (see Head and
Mayer, 2014 for a meta-analysis of the effect of RTAs on trade). In column
(2), we account for heterogeneity of firms and the mechanism of selection
of firms into exporters. The results remain significant, while the RTA
coefficient has higher value. Variables controlling for firm level
heterogeneity are all significant and of expected sign and magnitude (see
Table 2 in Helpman et al., 2008). Finally, in column (3) we present results
of IV estimation, which noticeably increases the estimates of the effects of
the policy variables, RTA in particular. This effect is well-documented in the
literature, which finds that not accounting for endogeneity of trade policy
considerably biases the estimates of trade policy variables downwards
(Baier and Bergstrand, 2007). In fact, our estimate of the RTA coefficient is
remarkably close to the estimate in Felbermayr et al. (2014), which equals
1.21. We also estimate the gravity mode by the PPML method, which
produced smaller coefficients of the effect of RTAs and EU on trade. This
fact is also well-documented, as discussed in Head & Mayer (2014). In order
to account for a range of coefficient estimates for RTA and EU variables, we
perform a sensitivity analysis in the robustness checks section.
5.3. Tariff and non-tariff measures: comparison of FTA and CU elasticity of trade
We also report the effects of different types of RTAs – FTA vs CU – in column
(5). The effect of CU on trade is much stronger, which is consistent with the
24
fact that CU goes much deeper than a FTA and removes non-tariff barriers
to trade, while FTAs are mostly focused on reduction of applied tariffs. In
fact, once we control for the level of tariffs, the positive effect of RTAs
disappears, while the effect of the EU on trade remains strongly positive
and significant. We will use these results later when we discuss the
differences in the effects of tariffs vs non-tariff measures and contrast FTAs
and CUs.
5.4. Dynamic effect of RTA and hysteresis
Two important issues need to be addressed in order to understand the
dynamics of transition between steady states. First, little is known about the
dynamics of the impact of an RTA on bilateral trade flows. Does it boost
trade at the early stage or does it increase trade gradually? Second, even
less is known about the effect of withdrawal from a trade agreement. Is it
symmetric and opposite in size? Does the RTA effect immediately disappear
or linger over time?
We start our discussion with the second of these issues. There are
not many examples countries walking away from a trade agreement in
recent times. However, the study of Head, Mayers and Ries (2010) on the
dynamics of trade of former colonies with their metropoles provides us with
some insights.15 Its main finding is in favour of a strong hysteresis effect of
the RTAs. The decline in bilateral trade between the metropole and its
colony is gradual, but very substantial. It drops to 35% of the initial level
within 30-40 years. Also, there is a similar decline in bilateral trade with the
other colonies of the same metropole. Finally, there is an overall decline in
15 Another example, which confirm this pattern, is the breakup of the Soviet Union, which led to dramatic reduction in bilateral trade between former USSR republics and to overall economic decline (Fidrmuc and Fidrmuc, 2003).
25
trade of about 20% after independence. One explanation of these patterns
may be that the metropole has trade policy capacity and expertise that is
commonly shared with its colonies. Once the colony gets independence, it
loses access to this facility, which has negative effect on its trade with all
countries.
To answer the question about the dynamics of the RTA effect, we
perform the analysis, similar to Head, Mayer, and Reis (2010). For each
bilateral trade agreement, we establish its starting point and include the
interaction terms of the RTA variable with its duration. Duration is
categorized from 1 to 40 years.16 We do the same for our EU variable. As
an alternative specification, we also consider whether the effect of an RTA
varies over time, by interacting the RTA variable with decade indicators.
The average effects on trade for RTA/EU members relative to non-members,
in percent, are presented in Figure 3.
Several important regularities emerge. First, the positive effect of
RTAs accumulates over time with some acceleration for RTAs over 25 years
old. Second, both RTA and, in particular, EU are currently having the
strongest impact on trade over the period 1960-2014. It follows from the
first effect and recent surge in RTA formation, but may also reinforce itself
as more and more countries rely on RTAs.
16 We include all RTAs with duration above 40 years in the same category.
26
Figure 3: Dynamics of RTA and EU in 1960-2014
020
040
060
0
Aver
age
effe
ct o
n tra
de w
ith o
ther
EU
coun
tries
, %
5 10 15 20 25 30 35 40Years since joining EU
010
020
030
040
0Av
erag
e ef
fect
of R
TA, %
5 10 15 20 25 30 35 40Years since forming RTA
-50
050
100
150
Aver
age
effe
ct o
n tra
de w
ithot
her E
U co
untri
es, %
1965197019751980198519901995200020052010Year
-20
020
4060
80Av
erag
e ef
fect
of R
TA, %
1965197019751980198519901995200020052010Year
We repeat the same exercise to study FTA and CU agreements separately.
CUs have the same pattern as RTAs in general (see Figure 4). However,
FTAs have a much weaker and shorter trade creation effect, which reaches
its maximum after 12 years from its formation and rapidly declines after
that. FTAs had the strongest trade creation effect in the 00’s, and are
currently experience a downward trend, while the trade creating effect of
the CUs accelerates and is currently at its peak.
27
Figure 4: Dynamics of FTA vs CU
010
020
030
040
0Av
erag
e ef
fect
of C
U, %
5 10 15 20 25 30 35 40Years since forming CU
020
4060
80Av
erag
e ef
fect
of F
TA, %
5 10 15 20 25 30 35 40Years since forming FTA
5010
015
020
025
0Av
erag
e ef
fect
of C
U, %
1965197019751980198519901995200020052010Year
020
4060
80Av
erag
e ef
fect
of F
TA, %
1965197019751980198519901995200020052010Year
5.5. Counterfactual scenarios: trade flows and welfare gains
As the benchmark, we take the estimated effects of the trade policy from
column (2) of Table 2 as given and estimate the effects of changes in trade
policy of the UK on its exports and imports, following the constrained PPML
methodology outlined in the previous section. The benchmark estimation
and simulations are performed using a 2012 sample; while we would like to
use most recent data to evaluate the effects of policy changes,
unfortunately the data for 2013 and 2014 has gaps. In what follows, we
present predictions and welfare gains based on the PPML simulation
exercise.
28
Table 3: UK actual exports, imports, and counterfactual changes by regions in 2012
Region Status Quo, TTIP & TTIP &No TTIP US UK
CMWTH
bln. USD EU, %Brexit, %
Brexit, % FTA, % FTA, %
A. Export from UKEU 230.35 -5.9 -38.1 -36.7 -39.7 -37.1US 45.76 46.8 -0.3 7.8 68.3 3.7Rest of the World 132.2 -10.1 3.1 6.8 -1.4 23.2All countries 408.31
B. Import to UKEU 347.7 -4.8 -33.8 -31.7 -33.6 -33.0US 54.86 54 9.3 14.1 80.7 7.9Rest of the World 257.03 -6.4 10.3 12.3 6.2 24.9All countries 659.59
Notes: Table reports actual exports and imports of UK in 2012 in USD to different groups of countries and counterfactual changes relative to the status quo levels in percent. We assume that a “shallow” FTA is in place.
Panel A of Table 3 presents changes in exports under the five scenarios.
The first column presents the status quo exports of the UK to EU, US and
the Rest of the World in 2012. The other five columns report percentage
changes from the status quo levels of exports for the five specified
scenarios. Relative to the status quo, signing the TTIP and remaining in the
EU would increase British exports to the US by 46.8 percent (20.9 billion
USD). It also would lower exports to the EU by 5.9 percent (13.5 billion USD)
and the rest of the world by 10.1 percent (13.3 billion USD). This result
illustrates the trade diversion caused by lowering trade barriers towards UK
goods sold in the US. Brexit combined with free trade between the US and
EU would lower exports from the UK to the EU by 38.1 percent, while the UK
exports to the US will remain almost unchanged.17 Furthermore, UK exports
to the rest of the World would increase by 3.1 percent. If the TTIP is not
17 Teulings (2017), who studied how Brexit influences exports found similar results.
29
signed and the UK still leaves the EU, this would cause a similar drop of
exports to the EU, and increase in exports to the US and Rest of the World
by 7.8 and 6.8 percent respectively. If the UK negotiates an FTA with the US
after Brexit, it would boost its exports to the UK by 68.3 percent. In case of
FTAs with 6 large Commonwealth countries, the UK expands its exports to
the rest of the world by 23.2 percent.
Panel B of Table 3 presents changes in British imports under the five
scenarios. These results are quite similar to the results for the UK exports
since we consider reciprocal trade liberalization; hence, trade barriers are
reduced identically for both exports and imports. Relative to the status quo,
signing the TTIP and remaining in the EU would increase British imports from
US by 54 percent. It also would lower imports from the EU by 4.8 percent or
by 16.7 billion USD. Brexit would lower UK imports from the EU by 32-34
percent or by 115-118 billion USD.
30
Table 4: Main welfare results
Elast. of subst. Welfare gains, %
TTIP EU TTIP
BrexitNo TTIP
and BrexitUS UK
FTACMWTH
FTA
A. Baseline results, “Shallow FTA EU/UK”, Table 2 column (2)
4 0.23 -3.97 -3.49 -2.66 -2.565 0.17 -2.99 -2.63 -2 -1.936 0.14 -2.4 -2.11 -1.61 -1.54Average 0.18 -3.12 -2.74 -2.09 -2.01
B. ‘Hard’ Brexit, Table 2 column (2)4 0.23 -6.71 -6.32 -5.47 -5.245 0.17 -5.08 -4.78 -4.13 -3.966 0.14 -4.08 -3.84 -3.32 -3.18Average 0.18 -5.29 -4.98 -4.31 -4.13
C. Only intensive margins, “Shallow FTA EU/UK”, Table 2 column (1)
4 0.17 -3.75 -3.46 -2.92 -2.815 0.13 -2.83 -2.61 -2.2 -2.116 0.1 -2.27 -2.09 -1.76 -1.69Average 0.13 -2.95 -2.72 -2.3 -2.2
D. IV results, “Shallow FTA EU/UK”, Table 2 column (3)4 0.85 -5.56 -3.84 -0.73 -0.725 0.63 -4.2 -2.9 -0.54 -0.546 0.51 -3.37 -2.32 -0.44 -0.43Average 0.66 -4.38 -3.02 -0.57 -0.56
Notes: Table presents a representative consumer welfare gains or losses due to 5 counterfactual scenarios. The gains and losses are measured relative to the status quo of the UK as a member of EU and no TTIP is signed. Computations are performed for different levels of sigma, which is elasticity of substitution parameter in the utility function.
Table 4 compares consumer welfare estimated according to (13) under
different scenarios and different levels of elasticity of substitution.
According to Head and Mayer (2014), 622 studies that used a structural
gravity approach found an average value of σ = 5.13. We have chosen three
levels of sigma: σ=4, σ=5 and σ=6. Panel A reports the results for RTA and
EU estimated effects taken from Table 2, column (2). It assumes a shallow
FTA is negotiated between the EU and UK, but NTMs between the EU and
31
UK will emerge. As expected, British consumers would experience the
largest losses in welfare of around 3.1 percent on average if the UK exits
the EU and the EU signs the TTIP. If the EU does not sign the TTIP, the impact
is slightly improved with a 2.74 percent loss. This result is similar to Dhingra
et al. (2017) who found a loss in welfare of 1.34 percent in the optimistic
case and 2.66 percent in the pessimistic case. Finally, signing the TTIP and
staying in the EU would increase consumer welfare in the UK by 0.17
percent. This result is lower than estimates of Felbermayr et al. (2015), but
similar to Francois et al. (2013). As results in columns (4) and (5) indicate,
the loss of welfare caused by Brexit cannot be compensated by signing an
FTA with the US or with Commonwealth countries. The losses of welfare in
those scenarios are still substantial, 2.1 and 2 percent, repectively.
Panel B of Table 4 present the welfare analysis of ‘hard’ Brexit, when
the UK withdraws from the single market, imposes MFN tariffs and both
sides erect NTM barriers typical to the ones they impose against countries
without preferential treatment. Such a negative scenario would lead to 4.1-
5.2 percent welfare losses relative to the benchmark status quo. Panel C of
Table 4 presents welfare changes if we ignore the effect of the extensive
margins on trade flows. As these results show, the trade-creating effect of
TTIP and potential FTAs is lower while the trade-reducing effect of Brexit is
lower. These results highlight importance of modelling both impact of trade
policy changes on extensive and intensive trade margins.
Finally, in Panel D of Table 4 we present the welfare analysis for the
trade effects estimated by the IV technique with point estimates taken from
column (3) of Table 2. Staying in the EU and signing TTIP would have
increased welfare in the UK by 0.66 percent. All Brexit scenarios generate
welfare losses. The losses from Brexit are stronger than the more
32
conservative estimates based on coefficients from column (3). Average loss
of welfare due to Brexit is 3 percent if TTIP is not signed and 4.4 percent if
TTIP is signed. Taking the elasticity of substitution equal to 4, conditional on
Brexit, British consumers would be almost 2 percent better off if the TTIP is
not signed. This result highlights benefits of being a member of a large
economic bloc in negotiating trade deals. Welfare losses of Brexit are
substantially mitigated by negotiating new trade deals with the US or
Commonwealth countries, leading to only 0.57 and 0.56 percent welfare
losses.
Finally, given estimation of the dynamic effects of RTA/EU elasticities of
trade presented in section 5.4, there are two conflicting tendencies that
should be taken into account when interpreting welfare gains presented in
Table 4. First, even the ‘hard’ Brexit scenario would not cause abrupt
welfare losses and catastrophic decline of trade flows between the EU and
UK. The changes occur gradually. On the other hand, currently, RTAs
especially CUs and EU in particular, have the strongest trade creation effect
over the last 50 years, which would indicate that the welfare losses of Brexit
are likely to be higher and welfare gains of TTIP would increase over time.
5.6. CU vs FTA
According to the comparison of the FTA and CU coefficients (column (5) of
Table 2), CUs generate 118% more trade then FTAs. This result is attributed
to the absence of non-tariff barriers and going through customs, associated
with CUs. Shallow FTAs, which target tariff reductions only, do not generate
much trade in the current low-tariff environment. Column (6) of Table 2
presents estimates of the elasticity of trade with respect to the applied tariff
while controlling for RTA and EU, which is close to -1. Given that the average
MFN rate in the EU is around 3 percent, this suggests that by keeping zero
33
applied tariffs with the EU, the UK would be able to avert an additional 3
percent reduction in its exports to the EU. However, the negative effect is
much larger. When we analyse the welfare impact, it does not translate into
noticeable welfare improvements for the scenarios that involve Brexit,
essentially leaving our baseline results intact.
6. Robustness checks
6.1. Resource cost barriers and welfare
We consider a sensitivity analysis exploring whether an increase/reduction
in resource costs, required to satisfy to NTM requirements (i.e. rules of
origin, technical standards, sanitary and phytosanitary measures)
associated with Brexit/TTIP, influences our welfare analysis. We model the
resource cost as additional cost that increases the price of output p' = p + τ
by certain value, , which is scaled in proportion to the trade costs . Exit τ ln φ
from the EU would increase resource costs required to organize border
controls, check NTM requirements and rule of origins, etc. TTIP, on the other
hand, would reduce resource costs due to NTM harmonization. Table 5
presents our results. Higher resource costs increase negative effect of
Brexit on welfare, while having a small effect on welfare gains for UK due to
TTIP. Welfare costs increase monotonically with increasing resource costs.
Table 5: Resource cost barriers and welfareAdditional resource Welfare gains, %cost, share oftotal trade costs TTIP EU TTIP Brexit No TTIP
and BrexitUS UK
FTACMWTH FTA
34
0.01 0.17 -3.00 -2.46 -1.83 -1.760.02 0.17 -3.31 -2.60 -1.97 -1.890.03 0.17 -3.62 -2.75 -2.11 -2.030.04 0.17 -3.93 -2.90 -2.24 -2.160.05 0.16 -4.23 -3.04 -2.38 -2.300.06 0.16 -4.54 -3.19 -2.52 -2.430.07 0.16 -4.85 -3.33 -2.66 -2.560.08 0.16 -5.15 -3.48 -2.79 -2.690.09 0.15 -5.45 -3.62 -2.93 -2.830.1 0.15 -5.75 -3.77 -3.07 -2.96
Notes: This table presents a representative consumer welfare gains or losses due to 5 counterfactual scenarios, under assumptions of soft Brexit, which maintains shallow FTA, but imposes NTMs on EU and the UK. The gains and losses are measured relative to the status quo of the UK as a member of EU and no TTIP is signed. Computations are performed for σ = 5.16
6.2. Sensitivity analysis: different levels of RTA impact on trade
Proponents of the TTIP may argue that our method of modelling this
arrangement as an average trade agreement, representative of past
agreements, is too conservative. They may argue that the TTIP is a
completely new, XXIst century type of FTA, which will go much further in
terms of the liberalization of goods, services, and removal of NTBs. On the
other hand, critics of the TTIP may claim that the effect of the agreement
will be close to zero because trade has already been liberalised between
the EU and US.
Table 6: Welfare gains for different levels of RTA impact
Scenario and RTA strength
Mean welfare gain, %
Standard Deviation
5th percentile,
%
95th percentile,
%
EU&TTIP, WCE1 0.15 0.21 -0.06 0.54
35
Brexit&TTIP, WCE2 -2.81 0.60 -3.87 -2.14
Brexit&No TTIP,
WCE3 -2.48 0.25 -2.92 -2.17
US UK FTA, WCE4 -1.91 0.45 -2.29 -1.14
CMWTH FTA, WCE5 -1.88 0.45 -2.36 -1.09
Notes: Computations are based on the estimated effect of EU from column (2) of Table 2. The effect of RTA is drawn from a distribution with mean 0.36 and standard deviation 0.42, which matches the sample moments for the distribution of the RTA coefficients from the meta study by Head and Mayer (2014). This analysis is repeated 100 times and the distribution of welfare gains is generated. Elasticity sigma is fixed at 5.13 for this exercise.
To confirm the robustness of our results for different levels of the TTIP
impact on trade flows we perform additional simulations. We consider a
distribution of plausible TTIP effects as given by the RTA coefficients from
the meta-study conducted by Head and Mayer (2014). They found that the
distribution of the impact of RTAs on trade flows, from 108 studies using
structurally estimated gravity modelling, to have a mean of 0.36 and
standard deviation of 0.42. Therefore, we model the potential TTIP impact
as a normally distributed random variable, . We take 100 N(0.36, 0.42)
random draws from this distribution and evaluate welfare gains for each
RTA draw, ranging from 0.31 to 1.76. The results of the distribution of
welfare gains for different policy scenarios and the elasticity of substitution
parameter are presented in Table 6. On average, Scenario 1 σ = 5.13
generates a welfare gain of 0.15 percent, Scenario 2 generates a welfare
loss of 2.81 percent, and Scenario 3 generates a welfare loss of 2.48
percent. Scenario 2 also generates the highest volatility. Scenarios 4 and 5
generate losses of 1.91 and 1.88 percent respectively. We also report 5th
and 95th percentiles of the distributions of gains for each scenario and
indicate them in the table. The results demonstrate that our welfare gain
estimates are quite robust to the error in the estimation of the effect of FTA.
Figure 5 presents the welfare gains depending on the strength of RTA
36
impact. These simulations indicate that the best option for the UK would be
to stay within the EU regardless of whether TTIP is signed or not. Moreover,
the negative effect of Brexit on UK welfare increases as the TTIP impact on
trade becomes stronger.
Figure 5: Welfare gains for different levels of RTA impact
-6-4
-20
2W
elfa
re g
ains
, %
-.5 0 .5 1 1.5 2RTA coefficient
EU&TTIP Brexit&TTIPBrexit US UK FTACMWTH FTA
6.3. Sensitivity analysis: TTIP and economic growth
For robustness, we also consider how the additional effects from the TTIP
would influence our results. Suppose the TTIP generates economic growth
in the country-members due to liberalization of capital and migration flows,
and reduction of NTBs. We model this as a uniform increase in GDP of the
TTIP members of percent, where takes values of 1 to 10 percent. Further, g g
we re-estimate our structural model and compute corresponding welfare
gains, assuming . The results are presented in Figure 6. As σ = 5.13
expected, the UK would benefit from being a member of TTIP, but if the UK
37
is out of EU, the additional indirect welfare gains from TTIP would not
compensate for the welfare losses from exiting the EU.
Figure 6: Welfare gains for different levels of GDP growth-1
2-8
-40
4W
elfa
re g
ains
, %
0 2 4 6 8 10GDP growth due to TTIP, %
EU&TTIP Brexit&TTIPBrexit
7. ConclusionGiven the current difficulties in achieving multilateral trade liberalization
alongside the geopolitical power struggle, our analysis indicates that
countries benefit from integrating into large regional trade/economic blocs.
Our results demonstrate that countries left out of this process experience
significant economic losses, stemming from trade diversion and trade
preference erosion. Our structural gravity results are benchmarked against
the findings in the existing literature. Furthermore, we have conducted
additional experiments to confirm the robustness of our results. This
emerging research framework provides a tractable model of global trade
without compromising its general equilibrium features. We estimate a lower
bound of UK welfare losses from exiting the EU and signing the TTIP. The
38
losses are derived from changes in trade costs, while output and
expenditure stay constant. If we account for other channels mentioned in
the literature, the negative effects of exiting the EU will be higher.
Depreciation of the pound after the Brexit referendum boosted
exports, but the effect is likely to be short lived. Cover and Mallick (2012)
highlight that historically exchange rate shocks have not had a significant
impact on UK business cycles. Moreover, consumers have already
experienced welfare losses due to the acceleration of inflation. Therefore,
the net overall effect of the pound depreciation is likely to be small in the
long run.
A resurgence of the Commonwealth group as a mega-regional would
be beneficial, although not to the extent that it would compensate for the
losses associated with Brexit. Furthermore, signing FTAs with the largest
commonwealth countries will not be without its challenges, in terms of
competing trade policy objectives as well as the perception of imbalanced
colonial-style partnerships. A US-UK FTA also faces significant political
difficulties and if realised, the benefits would not outweigh the negative
effect of Brexit.
The UK decision to step outside the EU framework and embark on
independently negotiating trade deals is a risky endeavour and likely to
bring significant welfare losses. The signature of the TTIP would further
weaken the UK’s position. The UK would be wise to negotiate a route to play
the new mega-regional ‘game’ as part of the EU bloc, although this seems
unlikely. It remains to be seen whether the UK can remain an integral part
of the EU project, perhaps even EU member in all but name, while seeming
to accept the democratic result of the referendum. Moreover, the UK’s lack
of commitment towards the EU project needs to reach some degree of
39
resolution so that members can focus on reforms and address the
geopolitical pressures from Asia.
We recognize that the TTIP can generate several important economic
benefits, which go beyond the scope of this paper. First, the TTIP can boost
productivity growth due to better access to a larger variety of inputs and
services. Second, it can generate efficiency gains due to external
economies of scale and scope. Third, it can boost knowledge spillovers, and
improve the allocation of capital and labour. Recent studies have introduced
dynamic knowledge accumulation into trade models and shown that the
dynamic welfare gains may be substantial (Sampson, 2016). Fourth, our
model does not look at sectoral effects of Brexit, which are important for
policy analysis. It is also true that introduction of sectors in the model may
change the welfare calculations, but only if one does not appropriately deal
with zero trade flows (Ossa, 2015). Our approach models and structurally
estimates selection into exporters with heterogeneous firms and the
possibility of zero trade flows. Finally, special treatment should be
considered for agricultural and services sectors. We leave these important
questions for future research.
References
Akhtar, S.I. and Jones, V.C., 2014, February. Transatlantic Trade and
Investment Partnership (TTIP) Negotiations. Library of Congress,
Congressional Research Service.
40
Anderson, J.E., Larch, M. and Yotov, Y.V., 2015. Growth and Trade with
Frictions: A Structural Estimation Framework (No. w21377). National
Bureau of Economic Research.
Anderson, K. and Norheim, H., 1993. From imperial to regional trade
preferences: Its effect on Europe’s intra-and extra-regional trade.
Weltwirtschaftliches Archiv, 129(1), pp.78-102.
Anderson, J.E. and Yotov, Y.V., 2010. The Changing Incidence of
Geography. American Economic Review, 100(5), pp.2157-86.
Anderton, C.H. and Carter, J.R., 2001. The impact of war on trade: An
interrupted times-series study. Journal of Peace Research, 38(4), pp.445-
457.
Arkolakis, C., Costinot, A. and Rodríguez-Clare, A., 2012. New Trade
Models, Same Old Gains? American Economic Review, 102(1), pp.94-130.
Armington, P.S., 1969. A theory of demand for products distinguished by
place of production. International Monetary Fund Staff Papers, 16(1),
pp.159-178.
Azevêdo, R. Embracing change: forging global trade partnerships.
Remarks by Director-General WTO. Inter-Pacific Bar Association, Kuala
Lumpur, plenary session, 14/04/16
41
Baier, S.L. and Bergstrand, J.H., 2007. Do free trade agreements actually
increase members' international trade?. Journal of International
Economics, 71(1), pp.72-95.
Baier, S. L., and Bergstrand, J. H. (2009). Estimating the effects of free
trade agreements on international trade flows using matching
econometrics. Journal of International Economics, 77(1), 63–76.
Baldwin, R. E., Francois, J. F., and Portes, R. (1997). The costs and benefits
of eastern enlargement: the impact on the EU and central Europe.
Economic Policy, 12(24), 125–176.
Baldwin, R. E., and Venables, A. J. (1995). Chapter 31 Regional economic
integration. Handbook of International Economics, 3(C), 1597–1644.
Berden, K. G., Francois, J., Thelle, M., Wymenga, P., & Tammines, S.
(2009). Non-Tariff Measures in EU-US Trade and Investment – An Economic
Analysis. Rotterdam: ECORYS Nederland BV, (December), 9–32.
Cover, J.P. and Mallick, S.K., 2012. Identifying sources of macroeconomic
and exchange rate fluctuations in the UK. Journal of International Money
and Finance, 31(6), pp.1627-1648.
Costinot, A. and Rodriguez-Clare, A., 2014. Trade Theory with Numbers:
Quantifying the Consequences of Globalization,” in Gita Gopinath, Elhanan
Helpman, and Kenneth Rogoff, eds., Handbook of International Economics,
Vol. 4 2014, chapter 4, pp. 197–261.
42
Dhingra, S., Huang, H., Ottaviano, G., Pessoa, J. P., Sampson, T., & Van
Reenen, J. (2017). The Costs and Benefits of Leaving the EU : Trade
Effects (CEP Discussion Paper No. 1478).
Dhingra, Swati, Gianmarco Ottaviano, Thomas Sampson, and van Reenen,
J. 2016. The Impact of Brexit on Foreign Investment in the UK (CEP Brexit
Analysis 3, Centre for Economic Performance 2016).
Disdier, A., Emlinger, C., and Fouré, J. 2016. Interdependencies between
Atlantic and Pacific agreements: Evidence from agri-food sectors.
Economic Modelling, 55, 241–253.
Ebell, M., Hurst, I., and Warren, J. 2016. Modelling the long-run economic
impact of leaving the European Union. Economic Modelling, 59, 196–209.
Eichengreen, B. and Irwin, D.A., 1998. The role of history in bilateral trade
flows. In The regionalization of the world economy (pp. 33-62). University
of Chicago Press.
Egger, P., Francois, J., Manchin, M. and Nelson, D., 2015. Non-tariff
barriers, integration and the transatlantic economy. Economic Policy,
30(83), pp.539-584.
Egger, Peter H. and Mario Larch. 2008. Interdependent Preferential Trade
Agreement Memberships: An Empirical Analysis, Journal of International
Economics 76(2), pp. 384-399.
43
Egger, P., Larch, M., Staub, K. E., and Winkelmann, R. 2011. The Trade
Effects of Endogenous Preferential Trade Agreements. American Economic
Journal: Economic Policy, 3(3), 113–143.
https://doi.org/10.1257/pol.3.3.113
Fally, T., 2015. Structural gravity and fixed effects. Journal of International
Economics, 97(1), pp.76-85.
Fan, Y., Ren, S., Cai, H. and Cui, X., 2014. The state's role and position in
international trade: A complex network perspective. Economic Modelling,
39, pp.71-81.
Felbermayr, G., Heid, B., Larch, M. and Yalcin, E., 2015. Macroeconomic
potentials of transatlantic free trade: a high resolution perspective for
Europe and the world. Economic Policy, 30(83), pp.491-537.
Fidrmuc, J., and Fidrmuc, J. 2003. Disintegration and trade. Review of
International Economics, 11(5), 811–829.
Francois, J., Manchin, M., Norberg, H., Pindyuk, O. and Tomberger, P.,
2013. Reducing transatlantic barriers to trade and investment: An
economic assessment. (No. 1503). Working Paper, Department of Economics,
Johannes Kepler University of Linz.
Froman, M., 2015. We are at the end of the line on the Doha Round of
trade talks. Financial Times, 13 December.
44
Guimaraes P, Portugal P., 2010. A simple feasible procedure to fit models
with high-dimensional fixed effects. Stata Journal. 10(4):628.
Head, K. and Mayer, T., 2014. Gravity Equations: Workhorse, Toolkit, and
Cookbook, Ch. 3 in Handbook of International Economics, Gopinath, G, E.
Helpman and K. Rogoff (Eds), Vol. 4, 131-95.
Head, K., Mayer, T. and Ries, J., 2010. The erosion of colonial trade
linkages after independence. Journal of International Economics, 81(1),
pp.1-14.
Helpman, E., Melitz, M. and Rubinstein, Y., 2008. Estimating trade flows:
trading partners and trading volumes. Quarterly Journal of Economics,
123(2), pp.441-487.
Hull, C., 1948. The Memoirs of Cordell Hull. Macmillan.
Irwin, D.A., 2011. Trade policy disaster: Lessons from the 1930s. MIT
press.
Krasner, S.D., 1976. State power and the structure of international trade.
World Politics, 28(03), pp.317-347.
Krugman, P., 1980. Scale economies, product differentiation, and the
pattern of trade. The American Economic Review, 70(5), pp.950-959.
45
Lake, J., and Yildiz, H. M. 2016. On the different geographic characteristics
of Free Trade Agreements and Customs Unions. Journal of International
Economics, 103, 213–233.
Lee, J.M., 1977. ‘Forward thinking’ and war: The colonial office during the
1940s. The Journal of Imperial and Commonwealth History, 6(1), pp.64-79.
Li, C., Wang, J. and Whalley, J., 2016. Impact of mega trade deals on
China: A computational general equilibrium analysis. Economic Modelling,
57, pp.13-25.
Low, P., 2015. TTIP and the world trading system. Catalyst? TTIP’s impact
on the Rest, pp.13-17.
Mavroidis, P.C., Irwin, D.A. and Sykes, A.O., 2009. The Genesis of the
GATT. Cambridge University Press.
Malmström, C., 2015. Transparency in TTIP. European Commission Blog
Post. https://ec.europa.eu/commission/2014-
2019/malmstrom/blog/transparency-ttip_en
Melitz, M.J., 2003. The impact of trade on intra‐industry reallocations and
aggregate industry productivity. Econometrica, 71(6), pp.1695-1725.
Melitz, J., and Toubal, F., 2014. Native Language, Spoken Language,
Translation and Trade. Journal of International Economics, 92(2), pp. 351-
363.
46
Pitlik, H., 2016. Who Disapproves of TTIP? Multiple Distrust in Companies
and Political Institutions (No. 513). WIFO.
Ossa, R., 2015. Why trade matters after all. Journal of International
Economics, 97(2), pp.266-277.
Ottaviano, G., Pessoa, J.P., Sampson, T. and van Reenen, J., 2014. Brexit or
Fixit? The trade and welfare effects of leaving the European Union. London
School of Economics and Political Science, LSE Library.
Sampson, T., 2016. Dynamic selection: An idea flows theory of entry,
trade, and growth. The Quarterly Journal of Economics, 131(1), pp.315-
380.
Sampson, T., 2017. Brexit: The economics of international
disintegration. Journal of Economic Perspectives, 31(4), pp.163-84.
Schott, J.J., Can Mega-Regionals Support Multilateralism?. OECD Global
Forum on Trade" Reconciling Regionalism and Multilateralism in a Post-Bali
World" Paris, 11 February 2014 (3).
Silva, J.S. and Tenreyro, S., 2006. The log of gravity. The Review of
Economics and statistics, 88(4), pp.641-658.
47
Teulings, R., 2017. Brexit and the Impact of Gradual Economic Integration
on Export. Tinbergen Institute Discussion Paper No. 17-075/VI. Available at
SSRN: https://ssrn.com/abstract=3021789
Toye, R., 2003. The Attlee government, the imperial preference system
and the creation of the GATT. The English Historical Review, 118(478),
pp.912-939.
Wilkinson, R., 2002. The World Trade Organization. New Political Economy,
7(1), pp.129-141.